ASCLDec 22, 2023

BLSTM-Based Confidence Estimation for End-to-End Speech Recognition

arXiv:2312.14609v16 citationsh-index: 41ICASSP
Originality Incremental advance
AI Analysis

This work improves confidence estimation for ASR applications, but it is incremental as it builds on existing methods for handling imbalanced data.

The paper tackles confidence estimation in end-to-end speech recognition by addressing the imbalanced dataset problem of detecting infrequent incorrect tokens in mostly correct sequences, achieving high performance with a BLSTM-based model that outperforms Transformer-based models.

Confidence estimation, in which we estimate the reliability of each recognized token (e.g., word, sub-word, and character) in automatic speech recognition (ASR) hypotheses and detect incorrectly recognized tokens, is an important function for developing ASR applications. In this study, we perform confidence estimation for end-to-end (E2E) ASR hypotheses. Recent E2E ASR systems show high performance (e.g., around 5% token error rates) for various ASR tasks. In such situations, confidence estimation becomes difficult since we need to detect infrequent incorrect tokens from mostly correct token sequences. To tackle this imbalanced dataset problem, we employ a bidirectional long short-term memory (BLSTM)-based model as a strong binary-class (correct/incorrect) sequence labeler that is trained with a class balancing objective. We experimentally confirmed that, by utilizing several types of ASR decoding scores as its auxiliary features, the model steadily shows high confidence estimation performance under highly imbalanced settings. We also confirmed that the BLSTM-based model outperforms Transformer-based confidence estimation models, which greatly underestimate incorrect tokens.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes